Efficient Processing Of Set Queries Using Bitmap Index

Growing complexity of enterprise-wide data and business processes necessitates the efficiency of complex decision support set queries. However, contemporary DBMS remain unsuccessful in handling set queries efficiently. In this thesis we propose efficient set query processing methods using bitmap index. The methods use bitmap vectors to represent attributes values in binary format. The methods test groups within a schema in a hierarchical fashion. Satisfying groups are bisected further and checked recursively while non-satisfying groups are pruned resulting in significant reduction in response times. In addition, our iterative implementation avoids the inefficiency that can be introduced by recursive implementation by reading the same bitmap vector for intersection many times. We also introduce pre-processing methods to reduce the complexity of the bitmap vectors, thus to improve the efficiency. Our implementation is based on FastBit, an open-source efficient compressed bitmap index framework. Experimental results on large datasets and comparison with results from PostgreSQL prove that our approach is superior owing to the fact that we are able to discard non-satisfying groups and capably optimize complex queries.